freebsd-dev/sys/kern/kern_prot.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

2495 lines
58 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: BSD-3-Clause
*
1994-05-24 10:09:53 +00:00
* Copyright (c) 1982, 1986, 1989, 1990, 1991, 1993
* The Regents of the University of California.
1994-05-24 10:09:53 +00:00
* (c) UNIX System Laboratories, Inc.
* Copyright (c) 2000-2001 Robert N. M. Watson.
* All rights reserved.
*
1994-05-24 10:09:53 +00:00
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the University nor the names of its contributors
1994-05-24 10:09:53 +00:00
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)kern_prot.c 8.6 (Berkeley) 1/21/94
*/
/*
* System calls related to processes and protection
*/
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_inet.h"
#include "opt_inet6.h"
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/acct.h>
#include <sys/kdb.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/loginclass.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/ptrace.h>
#include <sys/refcount.h>
#include <sys/sx.h>
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
#include <sys/priv.h>
#include <sys/proc.h>
#ifdef COMPAT_43
#include <sys/sysent.h>
#endif
#include <sys/sysproto.h>
#include <sys/jail.h>
#include <sys/racct.h>
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
#include <sys/rctl.h>
#include <sys/resourcevar.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/syscallsubr.h>
#include <sys/sysctl.h>
1994-05-24 10:09:53 +00:00
#ifdef REGRESSION
FEATURE(regression,
2012-01-08 18:48:36 +00:00
"Kernel support for interfaces necessary for regression testing (SECURITY RISK!)");
#endif
#include <security/audit/audit.h>
#include <security/mac/mac_framework.h>
static MALLOC_DEFINE(M_CRED, "cred", "credentials");
SYSCTL_NODE(_security, OID_AUTO, bsd, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
"BSD security policy");
2020-11-14 19:22:02 +00:00
static void crfree_final(struct ucred *cr);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
static void crsetgroups_locked(struct ucred *cr, int ngrp,
gid_t *groups);
#ifndef _SYS_SYSPROTO_H_
struct getpid_args {
1994-05-24 10:09:53 +00:00
int dummy;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_getpid(struct thread *td, struct getpid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
1994-05-24 10:09:53 +00:00
td->td_retval[0] = p->p_pid;
#if defined(COMPAT_43)
if (SV_PROC_FLAG(p, SV_AOUT))
td->td_retval[1] = kern_getppid(td);
1994-05-24 10:09:53 +00:00
#endif
return (0);
}
#ifndef _SYS_SYSPROTO_H_
struct getppid_args {
int dummy;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_getppid(struct thread *td, struct getppid_args *uap)
{
td->td_retval[0] = kern_getppid(td);
return (0);
}
int
kern_getppid(struct thread *td)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
1994-05-24 10:09:53 +00:00
return (p->p_oppid);
1994-05-24 10:09:53 +00:00
}
/*
* Get process group ID; note that POSIX getpgrp takes no parameter.
Commit major SMP cleanups and move the BGL (big giant lock) in the syscall path inward. A system call may select whether it needs the MP lock or not (the default being that it does need it). A great deal of conditional SMP code for various deadended experiments has been removed. 'cil' and 'cml' have been removed entirely, and the locking around the cpl has been removed. The conditional separately-locked fast-interrupt code has been removed, meaning that interrupts must hold the CPL now (but they pretty much had to anyway). Another reason for doing this is that the original separate-lock for interrupts just doesn't apply to the interrupt thread mechanism being contemplated. Modifications to the cpl may now ONLY occur while holding the MP lock. For example, if an otherwise MP safe syscall needs to mess with the cpl, it must hold the MP lock for the duration and must (as usual) save/restore the cpl in a nested fashion. This is precursor work for the real meat coming later: avoiding having to hold the MP lock for common syscalls and I/O's and interrupt threads. It is expected that the spl mechanisms and new interrupt threading mechanisms will be able to run in tandem, allowing a slow piecemeal transition to occur. This patch should result in a moderate performance improvement due to the considerable amount of code that has been removed from the critical path, especially the simplification of the spl*() calls. The real performance gains will come later. Approved by: jkh Reviewed by: current, bde (exception.s) Some work taken from: luoqi's patch
2000-03-28 07:16:37 +00:00
*/
#ifndef _SYS_SYSPROTO_H_
struct getpgrp_args {
int dummy;
};
#endif
int
sys_getpgrp(struct thread *td, struct getpgrp_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
1994-05-24 10:09:53 +00:00
PROC_LOCK(p);
td->td_retval[0] = p->p_pgrp->pg_id;
PROC_UNLOCK(p);
1994-05-24 10:09:53 +00:00
return (0);
}
/* Get an arbitrary pid's process group id */
#ifndef _SYS_SYSPROTO_H_
struct getpgid_args {
pid_t pid;
};
#endif
int
sys_getpgid(struct thread *td, struct getpgid_args *uap)
{
struct proc *p;
int error;
if (uap->pid == 0) {
p = td->td_proc;
PROC_LOCK(p);
} else {
p = pfind(uap->pid);
if (p == NULL)
return (ESRCH);
error = p_cansee(td, p);
if (error) {
PROC_UNLOCK(p);
return (error);
}
}
td->td_retval[0] = p->p_pgrp->pg_id;
PROC_UNLOCK(p);
return (0);
}
/*
* Get an arbitrary pid's session id.
*/
#ifndef _SYS_SYSPROTO_H_
struct getsid_args {
pid_t pid;
};
#endif
int
sys_getsid(struct thread *td, struct getsid_args *uap)
{
return (kern_getsid(td, uap->pid));
}
int
kern_getsid(struct thread *td, pid_t pid)
{
struct proc *p;
int error;
if (pid == 0) {
p = td->td_proc;
PROC_LOCK(p);
} else {
p = pfind(pid);
if (p == NULL)
return (ESRCH);
error = p_cansee(td, p);
if (error) {
PROC_UNLOCK(p);
return (error);
}
}
td->td_retval[0] = p->p_session->s_sid;
PROC_UNLOCK(p);
return (0);
}
#ifndef _SYS_SYSPROTO_H_
struct getuid_args {
int dummy;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_getuid(struct thread *td, struct getuid_args *uap)
1994-05-24 10:09:53 +00:00
{
td->td_retval[0] = td->td_ucred->cr_ruid;
#if defined(COMPAT_43)
td->td_retval[1] = td->td_ucred->cr_uid;
1994-05-24 10:09:53 +00:00
#endif
return (0);
}
#ifndef _SYS_SYSPROTO_H_
struct geteuid_args {
int dummy;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_geteuid(struct thread *td, struct geteuid_args *uap)
1994-05-24 10:09:53 +00:00
{
td->td_retval[0] = td->td_ucred->cr_uid;
1994-05-24 10:09:53 +00:00
return (0);
}
#ifndef _SYS_SYSPROTO_H_
struct getgid_args {
int dummy;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_getgid(struct thread *td, struct getgid_args *uap)
1994-05-24 10:09:53 +00:00
{
td->td_retval[0] = td->td_ucred->cr_rgid;
#if defined(COMPAT_43)
td->td_retval[1] = td->td_ucred->cr_groups[0];
1994-05-24 10:09:53 +00:00
#endif
return (0);
}
/*
* Get effective group ID. The "egid" is groups[0], and could be obtained
* via getgroups. This syscall exists because it is somewhat painful to do
* correctly in a library function.
*/
#ifndef _SYS_SYSPROTO_H_
struct getegid_args {
int dummy;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_getegid(struct thread *td, struct getegid_args *uap)
1994-05-24 10:09:53 +00:00
{
td->td_retval[0] = td->td_ucred->cr_groups[0];
1994-05-24 10:09:53 +00:00
return (0);
}
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct getgroups_args {
int gidsetsize;
1994-05-24 10:09:53 +00:00
gid_t *gidset;
};
#endif
int
sys_getgroups(struct thread *td, struct getgroups_args *uap)
1994-05-24 10:09:53 +00:00
{
struct ucred *cred;
int ngrp, error;
1994-05-24 10:09:53 +00:00
cred = td->td_ucred;
ngrp = cred->cr_ngroups;
if (uap->gidsetsize == 0) {
error = 0;
goto out;
1994-05-24 10:09:53 +00:00
}
if (uap->gidsetsize < ngrp)
return (EINVAL);
error = copyout(cred->cr_groups, uap->gidset, ngrp * sizeof(gid_t));
out:
td->td_retval[0] = ngrp;
return (error);
1994-05-24 10:09:53 +00:00
}
#ifndef _SYS_SYSPROTO_H_
struct setsid_args {
int dummy;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setsid(struct thread *td, struct setsid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct pgrp *pgrp;
int error;
struct proc *p = td->td_proc;
struct pgrp *newpgrp;
struct session *newsess;
error = 0;
pgrp = NULL;
1994-05-24 10:09:53 +00:00
newpgrp = uma_zalloc(pgrp_zone, M_WAITOK);
newsess = malloc(sizeof(struct session), M_SESSION, M_WAITOK | M_ZERO);
sx_xlock(&proctree_lock);
if (p->p_pgid == p->p_pid || (pgrp = pgfind(p->p_pid)) != NULL) {
if (pgrp != NULL)
PGRP_UNLOCK(pgrp);
error = EPERM;
} else {
(void)enterpgrp(p, p->p_pid, newpgrp, newsess);
td->td_retval[0] = p->p_pid;
newpgrp = NULL;
newsess = NULL;
1994-05-24 10:09:53 +00:00
}
sx_xunlock(&proctree_lock);
uma_zfree(pgrp_zone, newpgrp);
free(newsess, M_SESSION);
return (error);
1994-05-24 10:09:53 +00:00
}
/*
* set process group (setpgid/old setpgrp)
*
* caller does setpgid(targpid, targpgid)
*
* pid must be caller or child of caller (ESRCH)
* if a child
* pid must be in same session (EPERM)
* pid can't have done an exec (EACCES)
* if pgid != pid
* there must exist some pid in same session having pgid (EPERM)
* pid must not be session leader (EPERM)
*/
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct setpgid_args {
int pid; /* target process id */
int pgid; /* target pgrp id */
1994-05-24 10:09:53 +00:00
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setpgid(struct thread *td, struct setpgid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *curp = td->td_proc;
struct proc *targp; /* target process */
struct pgrp *pgrp; /* target pgrp */
int error;
struct pgrp *newpgrp;
1994-05-24 10:09:53 +00:00
if (uap->pgid < 0)
return (EINVAL);
error = 0;
newpgrp = uma_zalloc(pgrp_zone, M_WAITOK);
sx_xlock(&proctree_lock);
1994-05-24 10:09:53 +00:00
if (uap->pid != 0 && uap->pid != curp->p_pid) {
if ((targp = pfind(uap->pid)) == NULL) {
error = ESRCH;
goto done;
}
if (!inferior(targp)) {
PROC_UNLOCK(targp);
error = ESRCH;
goto done;
}
if ((error = p_cansee(td, targp))) {
PROC_UNLOCK(targp);
goto done;
}
if (targp->p_pgrp == NULL ||
targp->p_session != curp->p_session) {
PROC_UNLOCK(targp);
error = EPERM;
goto done;
}
if (targp->p_flag & P_EXEC) {
PROC_UNLOCK(targp);
error = EACCES;
goto done;
}
PROC_UNLOCK(targp);
} else
1994-05-24 10:09:53 +00:00
targp = curp;
if (SESS_LEADER(targp)) {
error = EPERM;
goto done;
}
if (uap->pgid == 0)
1994-05-24 10:09:53 +00:00
uap->pgid = targp->p_pid;
if ((pgrp = pgfind(uap->pgid)) == NULL) {
if (uap->pgid == targp->p_pid) {
error = enterpgrp(targp, uap->pgid, newpgrp,
NULL);
if (error == 0)
newpgrp = NULL;
} else
error = EPERM;
} else {
if (pgrp == targp->p_pgrp) {
PGRP_UNLOCK(pgrp);
goto done;
}
if (pgrp->pg_id != targp->p_pid &&
pgrp->pg_session != curp->p_session) {
PGRP_UNLOCK(pgrp);
error = EPERM;
goto done;
}
PGRP_UNLOCK(pgrp);
error = enterthispgrp(targp, pgrp);
}
done:
sx_xunlock(&proctree_lock);
KASSERT((error == 0) || (newpgrp != NULL),
("setpgid failed and newpgrp is NULL"));
uma_zfree(pgrp_zone, newpgrp);
return (error);
1994-05-24 10:09:53 +00:00
}
/*
* Use the clause in B.4.2.2 that allows setuid/setgid to be 4.2/4.3BSD
* compatible. It says that setting the uid/gid to euid/egid is a special
* case of "appropriate privilege". Once the rules are expanded out, this
* basically means that setuid(nnn) sets all three id's, in all permitted
* cases unless _POSIX_SAVED_IDS is enabled. In that case, setuid(getuid())
* does not set the saved id - this is dangerous for traditional BSD
* programs. For this reason, we *really* do not want to set
* _POSIX_SAVED_IDS and do not want to clear POSIX_APPENDIX_B_4_2_2.
*/
#define POSIX_APPENDIX_B_4_2_2
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct setuid_args {
uid_t uid;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setuid(struct thread *td, struct setuid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
struct ucred *newcred, *oldcred;
uid_t uid;
struct uidinfo *uip;
int error;
1994-05-24 10:09:53 +00:00
uid = uap->uid;
AUDIT_ARG_UID(uid);
newcred = crget();
uip = uifind(uid);
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
/*
* Copy credentials so other references do not see our changes.
*/
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_setuid(oldcred, uid);
if (error)
goto fail;
#endif
/*
* See if we have "permission" by POSIX 1003.1 rules.
*
* Note that setuid(geteuid()) is a special case of
* "appropriate privileges" in appendix B.4.2.2. We need
* to use this clause to be compatible with traditional BSD
* semantics. Basically, it means that "setuid(xx)" sets all
* three id's (assuming you have privs).
*
* Notes on the logic. We do things in three steps.
* 1: We determine if the euid is going to change, and do EPERM
* right away. We unconditionally change the euid later if this
* test is satisfied, simplifying that part of the logic.
* 2: We determine if the real and/or saved uids are going to
* change. Determined by compile options.
* 3: Change euid last. (after tests in #2 for "appropriate privs")
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (uid != oldcred->cr_ruid && /* allow setuid(getuid()) */
#ifdef _POSIX_SAVED_IDS
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
uid != oldcred->cr_svuid && /* allow setuid(saved gid) */
#endif
#ifdef POSIX_APPENDIX_B_4_2_2 /* Use BSD-compat clause from B.4.2.2 */
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
uid != oldcred->cr_uid && /* allow setuid(geteuid()) */
#endif
(error = priv_check_cred(oldcred, PRIV_CRED_SETUID)) != 0)
goto fail;
#ifdef _POSIX_SAVED_IDS
1994-05-24 10:09:53 +00:00
/*
* Do we have "appropriate privileges" (are we root or uid == euid)
* If so, we are changing the real uid and/or saved uid.
1994-05-24 10:09:53 +00:00
*/
if (
#ifdef POSIX_APPENDIX_B_4_2_2 /* Use the clause from B.4.2.2 */
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
uid == oldcred->cr_uid ||
#endif
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/* We are using privs. */
priv_check_cred(oldcred, PRIV_CRED_SETUID) == 0)
#endif
{
/*
* Set the real uid and transfer proc count to new user.
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (uid != oldcred->cr_ruid) {
change_ruid(newcred, uip);
setsugid(p);
}
/*
* Set saved uid
*
* XXX always set saved uid even if not _POSIX_SAVED_IDS, as
* the security of seteuid() depends on it. B.4.2.2 says it
* is important that we should do this.
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (uid != oldcred->cr_svuid) {
change_svuid(newcred, uid);
setsugid(p);
}
}
/*
* In all permitted cases, we are changing the euid.
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (uid != oldcred->cr_uid) {
change_euid(newcred, uip);
setsugid(p);
}
proc_set_cred(p, newcred);
#ifdef RACCT
racct_proc_ucred_changed(p, oldcred, newcred);
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
crhold(newcred);
#endif
PROC_UNLOCK(p);
#ifdef RCTL
rctl_proc_ucred_changed(p, newcred);
crfree(newcred);
#endif
uifree(uip);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
uifree(uip);
crfree(newcred);
return (error);
1994-05-24 10:09:53 +00:00
}
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct seteuid_args {
uid_t euid;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_seteuid(struct thread *td, struct seteuid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
struct ucred *newcred, *oldcred;
uid_t euid;
struct uidinfo *euip;
int error;
1994-05-24 10:09:53 +00:00
euid = uap->euid;
AUDIT_ARG_EUID(euid);
newcred = crget();
euip = uifind(euid);
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
/*
* Copy credentials so other references do not see our changes.
*/
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_seteuid(oldcred, euid);
if (error)
goto fail;
#endif
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (euid != oldcred->cr_ruid && /* allow seteuid(getuid()) */
euid != oldcred->cr_svuid && /* allow seteuid(saved uid) */
(error = priv_check_cred(oldcred, PRIV_CRED_SETEUID)) != 0)
goto fail;
1994-05-24 10:09:53 +00:00
/*
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
* Everything's okay, do it.
1994-05-24 10:09:53 +00:00
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (oldcred->cr_uid != euid) {
change_euid(newcred, euip);
setsugid(p);
}
proc_set_cred(p, newcred);
PROC_UNLOCK(p);
uifree(euip);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
uifree(euip);
crfree(newcred);
return (error);
1994-05-24 10:09:53 +00:00
}
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct setgid_args {
gid_t gid;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setgid(struct thread *td, struct setgid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
struct ucred *newcred, *oldcred;
gid_t gid;
int error;
1994-05-24 10:09:53 +00:00
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
gid = uap->gid;
AUDIT_ARG_GID(gid);
newcred = crget();
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_setgid(oldcred, gid);
if (error)
goto fail;
#endif
/*
* See if we have "permission" by POSIX 1003.1 rules.
*
* Note that setgid(getegid()) is a special case of
* "appropriate privileges" in appendix B.4.2.2. We need
* to use this clause to be compatible with traditional BSD
* semantics. Basically, it means that "setgid(xx)" sets all
* three id's (assuming you have privs).
*
* For notes on the logic here, see setuid() above.
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (gid != oldcred->cr_rgid && /* allow setgid(getgid()) */
#ifdef _POSIX_SAVED_IDS
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
gid != oldcred->cr_svgid && /* allow setgid(saved gid) */
#endif
#ifdef POSIX_APPENDIX_B_4_2_2 /* Use BSD-compat clause from B.4.2.2 */
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
gid != oldcred->cr_groups[0] && /* allow setgid(getegid()) */
#endif
(error = priv_check_cred(oldcred, PRIV_CRED_SETGID)) != 0)
goto fail;
#ifdef _POSIX_SAVED_IDS
/*
* Do we have "appropriate privileges" (are we root or gid == egid)
* If so, we are changing the real uid and saved gid.
*/
if (
#ifdef POSIX_APPENDIX_B_4_2_2 /* use the clause from B.4.2.2 */
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
gid == oldcred->cr_groups[0] ||
#endif
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
/* We are using privs. */
priv_check_cred(oldcred, PRIV_CRED_SETGID) == 0)
#endif
{
/*
* Set real gid
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (oldcred->cr_rgid != gid) {
change_rgid(newcred, gid);
setsugid(p);
}
/*
* Set saved gid
*
* XXX always set saved gid even if not _POSIX_SAVED_IDS, as
* the security of setegid() depends on it. B.4.2.2 says it
* is important that we should do this.
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (oldcred->cr_svgid != gid) {
change_svgid(newcred, gid);
setsugid(p);
}
}
/*
* In all cases permitted cases, we are changing the egid.
* Copy credentials so other references do not see our changes.
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (oldcred->cr_groups[0] != gid) {
change_egid(newcred, gid);
setsugid(p);
}
proc_set_cred(p, newcred);
PROC_UNLOCK(p);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
crfree(newcred);
return (error);
1994-05-24 10:09:53 +00:00
}
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct setegid_args {
gid_t egid;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setegid(struct thread *td, struct setegid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
struct ucred *newcred, *oldcred;
gid_t egid;
int error;
1994-05-24 10:09:53 +00:00
egid = uap->egid;
AUDIT_ARG_EGID(egid);
newcred = crget();
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_setegid(oldcred, egid);
if (error)
goto fail;
#endif
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (egid != oldcred->cr_rgid && /* allow setegid(getgid()) */
egid != oldcred->cr_svgid && /* allow setegid(saved gid) */
(error = priv_check_cred(oldcred, PRIV_CRED_SETEGID)) != 0)
goto fail;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (oldcred->cr_groups[0] != egid) {
change_egid(newcred, egid);
setsugid(p);
}
proc_set_cred(p, newcred);
PROC_UNLOCK(p);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
crfree(newcred);
return (error);
1994-05-24 10:09:53 +00:00
}
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct setgroups_args {
int gidsetsize;
1994-05-24 10:09:53 +00:00
gid_t *gidset;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setgroups(struct thread *td, struct setgroups_args *uap)
{
gid_t smallgroups[XU_NGROUPS];
gid_t *groups;
int gidsetsize, error;
gidsetsize = uap->gidsetsize;
if (gidsetsize > ngroups_max + 1 || gidsetsize < 0)
return (EINVAL);
if (gidsetsize > XU_NGROUPS)
groups = malloc(gidsetsize * sizeof(gid_t), M_TEMP, M_WAITOK);
else
groups = smallgroups;
error = copyin(uap->gidset, groups, gidsetsize * sizeof(gid_t));
if (error == 0)
error = kern_setgroups(td, gidsetsize, groups);
if (gidsetsize > XU_NGROUPS)
free(groups, M_TEMP);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
return (error);
}
int
kern_setgroups(struct thread *td, u_int ngrp, gid_t *groups)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
struct ucred *newcred, *oldcred;
1994-05-24 10:09:53 +00:00
int error;
MPASS(ngrp <= ngroups_max + 1);
AUDIT_ARG_GROUPSET(groups, ngrp);
newcred = crget();
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
crextend(newcred, ngrp);
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_setgroups(oldcred, ngrp, groups);
if (error)
goto fail;
#endif
error = priv_check_cred(oldcred, PRIV_CRED_SETGROUPS);
if (error)
goto fail;
if (ngrp == 0) {
/*
* setgroups(0, NULL) is a legitimate way of clearing the
* groups vector on non-BSD systems (which generally do not
* have the egid in the groups[0]). We risk security holes
* when running non-BSD software if we do not do the same.
*/
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
newcred->cr_ngroups = 1;
} else {
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
crsetgroups_locked(newcred, ngrp, groups);
}
setsugid(p);
proc_set_cred(p, newcred);
PROC_UNLOCK(p);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
crfree(newcred);
return (error);
1994-05-24 10:09:53 +00:00
}
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct setreuid_args {
uid_t ruid;
uid_t euid;
1994-05-24 10:09:53 +00:00
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setreuid(struct thread *td, struct setreuid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
struct ucred *newcred, *oldcred;
uid_t euid, ruid;
struct uidinfo *euip, *ruip;
int error;
1994-05-24 10:09:53 +00:00
euid = uap->euid;
ruid = uap->ruid;
AUDIT_ARG_EUID(euid);
AUDIT_ARG_RUID(ruid);
newcred = crget();
euip = uifind(euid);
ruip = uifind(ruid);
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_setreuid(oldcred, ruid, euid);
if (error)
goto fail;
#endif
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (((ruid != (uid_t)-1 && ruid != oldcred->cr_ruid &&
ruid != oldcred->cr_svuid) ||
(euid != (uid_t)-1 && euid != oldcred->cr_uid &&
euid != oldcred->cr_ruid && euid != oldcred->cr_svuid)) &&
(error = priv_check_cred(oldcred, PRIV_CRED_SETREUID)) != 0)
goto fail;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (euid != (uid_t)-1 && oldcred->cr_uid != euid) {
change_euid(newcred, euip);
setsugid(p);
}
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (ruid != (uid_t)-1 && oldcred->cr_ruid != ruid) {
change_ruid(newcred, ruip);
setsugid(p);
}
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if ((ruid != (uid_t)-1 || newcred->cr_uid != newcred->cr_ruid) &&
newcred->cr_svuid != newcred->cr_uid) {
change_svuid(newcred, newcred->cr_uid);
setsugid(p);
}
proc_set_cred(p, newcred);
#ifdef RACCT
racct_proc_ucred_changed(p, oldcred, newcred);
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
crhold(newcred);
#endif
PROC_UNLOCK(p);
#ifdef RCTL
rctl_proc_ucred_changed(p, newcred);
crfree(newcred);
#endif
uifree(ruip);
uifree(euip);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
uifree(ruip);
uifree(euip);
crfree(newcred);
return (error);
1994-05-24 10:09:53 +00:00
}
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct setregid_args {
gid_t rgid;
gid_t egid;
1994-05-24 10:09:53 +00:00
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setregid(struct thread *td, struct setregid_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
struct ucred *newcred, *oldcred;
gid_t egid, rgid;
int error;
1994-05-24 10:09:53 +00:00
egid = uap->egid;
rgid = uap->rgid;
AUDIT_ARG_EGID(egid);
AUDIT_ARG_RGID(rgid);
newcred = crget();
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_setregid(oldcred, rgid, egid);
if (error)
goto fail;
#endif
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (((rgid != (gid_t)-1 && rgid != oldcred->cr_rgid &&
rgid != oldcred->cr_svgid) ||
(egid != (gid_t)-1 && egid != oldcred->cr_groups[0] &&
egid != oldcred->cr_rgid && egid != oldcred->cr_svgid)) &&
(error = priv_check_cred(oldcred, PRIV_CRED_SETREGID)) != 0)
goto fail;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (egid != (gid_t)-1 && oldcred->cr_groups[0] != egid) {
change_egid(newcred, egid);
setsugid(p);
}
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (rgid != (gid_t)-1 && oldcred->cr_rgid != rgid) {
change_rgid(newcred, rgid);
setsugid(p);
}
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if ((rgid != (gid_t)-1 || newcred->cr_groups[0] != newcred->cr_rgid) &&
newcred->cr_svgid != newcred->cr_groups[0]) {
change_svgid(newcred, newcred->cr_groups[0]);
setsugid(p);
}
proc_set_cred(p, newcred);
PROC_UNLOCK(p);
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
crfree(newcred);
return (error);
1994-05-24 10:09:53 +00:00
}
/*
* setresuid(ruid, euid, suid) is like setreuid except control over the saved
* uid is explicit.
*/
#ifndef _SYS_SYSPROTO_H_
struct setresuid_args {
uid_t ruid;
uid_t euid;
uid_t suid;
};
#endif
/* ARGSUSED */
int
sys_setresuid(struct thread *td, struct setresuid_args *uap)
{
struct proc *p = td->td_proc;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
struct ucred *newcred, *oldcred;
uid_t euid, ruid, suid;
struct uidinfo *euip, *ruip;
int error;
euid = uap->euid;
ruid = uap->ruid;
suid = uap->suid;
AUDIT_ARG_EUID(euid);
AUDIT_ARG_RUID(ruid);
AUDIT_ARG_SUID(suid);
newcred = crget();
euip = uifind(euid);
ruip = uifind(ruid);
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_setresuid(oldcred, ruid, euid, suid);
if (error)
goto fail;
#endif
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (((ruid != (uid_t)-1 && ruid != oldcred->cr_ruid &&
ruid != oldcred->cr_svuid &&
ruid != oldcred->cr_uid) ||
(euid != (uid_t)-1 && euid != oldcred->cr_ruid &&
euid != oldcred->cr_svuid &&
euid != oldcred->cr_uid) ||
(suid != (uid_t)-1 && suid != oldcred->cr_ruid &&
suid != oldcred->cr_svuid &&
suid != oldcred->cr_uid)) &&
(error = priv_check_cred(oldcred, PRIV_CRED_SETRESUID)) != 0)
goto fail;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (euid != (uid_t)-1 && oldcred->cr_uid != euid) {
change_euid(newcred, euip);
setsugid(p);
}
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (ruid != (uid_t)-1 && oldcred->cr_ruid != ruid) {
change_ruid(newcred, ruip);
setsugid(p);
}
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (suid != (uid_t)-1 && oldcred->cr_svuid != suid) {
change_svuid(newcred, suid);
setsugid(p);
}
proc_set_cred(p, newcred);
#ifdef RACCT
racct_proc_ucred_changed(p, oldcred, newcred);
call racct_proc_ucred_changed() under the proc lock The lock is required to ensure that the switch to the new credentials and the transfer of the process's accounting data from the old credentials to the new ones is done atomically. Otherwise, some updates may be applied to the new credentials and then additionally transferred from the old credentials if the updates happen after proc_set_cred() and before racct_proc_ucred_changed(). The problem is especially pronounced for RACCT_RSS because - there is a strict accounting for this resource (it's reclaimable) - it's updated asynchronously by the vm daemon - it's updated by setting an absolute value instead of applying a delta I had to remove a call to rctl_proc_ucred_changed() from racct_proc_ucred_changed() and make all callers of latter call the former as well. The reason is that rctl_proc_ucred_changed, as it is implemented now, cannot be called while holding the proc lock, so the lock is dropped after calling racct_proc_ucred_changed. Additionally, I've added calls to crhold / crfree around the rctl call, because without the proc lock there is no gurantee that the new credentials, owned by the process, will stay stable. That does not eliminate a possibility that the credentials passed to the rctl will get stale. Ideally, rctl_proc_ucred_changed should be able to work under the proc lock. Many thanks to kib for pointing out the above problems. PR: 222027 Discussed with: kib No comment: trasz MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D15048
2018-04-20 13:08:04 +00:00
crhold(newcred);
#endif
PROC_UNLOCK(p);
#ifdef RCTL
rctl_proc_ucred_changed(p, newcred);
crfree(newcred);
#endif
uifree(ruip);
uifree(euip);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
uifree(ruip);
uifree(euip);
crfree(newcred);
return (error);
}
/*
* setresgid(rgid, egid, sgid) is like setregid except control over the saved
* gid is explicit.
*/
#ifndef _SYS_SYSPROTO_H_
struct setresgid_args {
gid_t rgid;
gid_t egid;
gid_t sgid;
};
#endif
/* ARGSUSED */
int
sys_setresgid(struct thread *td, struct setresgid_args *uap)
{
struct proc *p = td->td_proc;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
struct ucred *newcred, *oldcred;
gid_t egid, rgid, sgid;
int error;
egid = uap->egid;
rgid = uap->rgid;
sgid = uap->sgid;
AUDIT_ARG_EGID(egid);
AUDIT_ARG_RGID(rgid);
AUDIT_ARG_SGID(sgid);
newcred = crget();
PROC_LOCK(p);
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
oldcred = crcopysafe(p, newcred);
#ifdef MAC
error = mac_cred_check_setresgid(oldcred, rgid, egid, sgid);
if (error)
goto fail;
#endif
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (((rgid != (gid_t)-1 && rgid != oldcred->cr_rgid &&
rgid != oldcred->cr_svgid &&
rgid != oldcred->cr_groups[0]) ||
(egid != (gid_t)-1 && egid != oldcred->cr_rgid &&
egid != oldcred->cr_svgid &&
egid != oldcred->cr_groups[0]) ||
(sgid != (gid_t)-1 && sgid != oldcred->cr_rgid &&
sgid != oldcred->cr_svgid &&
sgid != oldcred->cr_groups[0])) &&
(error = priv_check_cred(oldcred, PRIV_CRED_SETRESGID)) != 0)
goto fail;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (egid != (gid_t)-1 && oldcred->cr_groups[0] != egid) {
change_egid(newcred, egid);
setsugid(p);
}
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (rgid != (gid_t)-1 && oldcred->cr_rgid != rgid) {
change_rgid(newcred, rgid);
setsugid(p);
}
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
if (sgid != (gid_t)-1 && oldcred->cr_svgid != sgid) {
change_svgid(newcred, sgid);
setsugid(p);
}
proc_set_cred(p, newcred);
PROC_UNLOCK(p);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
crfree(oldcred);
return (0);
fail:
PROC_UNLOCK(p);
crfree(newcred);
return (error);
}
#ifndef _SYS_SYSPROTO_H_
struct getresuid_args {
uid_t *ruid;
uid_t *euid;
uid_t *suid;
};
#endif
/* ARGSUSED */
int
sys_getresuid(struct thread *td, struct getresuid_args *uap)
{
struct ucred *cred;
int error1 = 0, error2 = 0, error3 = 0;
cred = td->td_ucred;
if (uap->ruid)
error1 = copyout(&cred->cr_ruid,
uap->ruid, sizeof(cred->cr_ruid));
if (uap->euid)
error2 = copyout(&cred->cr_uid,
uap->euid, sizeof(cred->cr_uid));
if (uap->suid)
error3 = copyout(&cred->cr_svuid,
uap->suid, sizeof(cred->cr_svuid));
return (error1 ? error1 : error2 ? error2 : error3);
}
#ifndef _SYS_SYSPROTO_H_
struct getresgid_args {
gid_t *rgid;
gid_t *egid;
gid_t *sgid;
};
#endif
/* ARGSUSED */
int
sys_getresgid(struct thread *td, struct getresgid_args *uap)
{
struct ucred *cred;
int error1 = 0, error2 = 0, error3 = 0;
cred = td->td_ucred;
if (uap->rgid)
error1 = copyout(&cred->cr_rgid,
uap->rgid, sizeof(cred->cr_rgid));
if (uap->egid)
error2 = copyout(&cred->cr_groups[0],
uap->egid, sizeof(cred->cr_groups[0]));
if (uap->sgid)
error3 = copyout(&cred->cr_svgid,
uap->sgid, sizeof(cred->cr_svgid));
return (error1 ? error1 : error2 ? error2 : error3);
}
#ifndef _SYS_SYSPROTO_H_
struct issetugid_args {
int dummy;
};
#endif
/* ARGSUSED */
int
sys_issetugid(struct thread *td, struct issetugid_args *uap)
{
struct proc *p = td->td_proc;
/*
* Note: OpenBSD sets a P_SUGIDEXEC flag set at execve() time,
* we use P_SUGID because we consider changing the owners as
* "tainting" as well.
* This is significant for procs that start as root and "become"
* a user without an exec - programs cannot know *everything*
* that libc *might* have put in their data segment.
*/
td->td_retval[0] = (p->p_flag & P_SUGID) ? 1 : 0;
return (0);
}
int
sys___setugid(struct thread *td, struct __setugid_args *uap)
{
#ifdef REGRESSION
struct proc *p;
p = td->td_proc;
switch (uap->flag) {
case 0:
PROC_LOCK(p);
p->p_flag &= ~P_SUGID;
PROC_UNLOCK(p);
return (0);
case 1:
PROC_LOCK(p);
p->p_flag |= P_SUGID;
PROC_UNLOCK(p);
return (0);
default:
return (EINVAL);
}
#else /* !REGRESSION */
return (ENOSYS);
#endif /* REGRESSION */
}
1994-05-24 10:09:53 +00:00
/*
* Check if gid is a member of the group set.
*/
int
groupmember(gid_t gid, struct ucred *cred)
1994-05-24 10:09:53 +00:00
{
int l;
int h;
int m;
if (cred->cr_groups[0] == gid)
return(1);
/*
* If gid was not our primary group, perform a binary search
* of the supplemental groups. This is possible because we
* sort the groups in crsetgroups().
*/
l = 1;
h = cred->cr_ngroups;
while (l < h) {
m = l + ((h - l) / 2);
if (cred->cr_groups[m] < gid)
l = m + 1;
else
h = m;
}
if ((l < cred->cr_ngroups) && (cred->cr_groups[l] == gid))
return (1);
1994-05-24 10:09:53 +00:00
return (0);
}
/*
* Test the active securelevel against a given level. securelevel_gt()
* implements (securelevel > level). securelevel_ge() implements
* (securelevel >= level). Note that the logic is inverted -- these
* functions return EPERM on "success" and 0 on "failure".
*
* Due to care taken when setting the securelevel, we know that no jail will
* be less secure that its parent (or the physical system), so it is sufficient
* to test the current jail only.
*
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
* XXXRW: Possibly since this has to do with privilege, it should move to
* kern_priv.c.
*/
int
securelevel_gt(struct ucred *cr, int level)
{
return (cr->cr_prison->pr_securelevel > level ? EPERM : 0);
}
int
securelevel_ge(struct ucred *cr, int level)
{
return (cr->cr_prison->pr_securelevel >= level ? EPERM : 0);
}
/*
* 'see_other_uids' determines whether or not visibility of processes
* and sockets with credentials holding different real uids is possible
* using a variety of system MIBs.
* XXX: data declarations should be together near the beginning of the file.
*/
static int see_other_uids = 1;
SYSCTL_INT(_security_bsd, OID_AUTO, see_other_uids, CTLFLAG_RW,
&see_other_uids, 0,
"Unprivileged processes may see subjects/objects with different real uid");
/*-
* Determine if u1 "can see" the subject specified by u2, according to the
* 'see_other_uids' policy.
* Returns: 0 for permitted, ESRCH otherwise
* Locks: none
* References: *u1 and *u2 must not change during the call
* u1 may equal u2, in which case only one reference is required
*/
int
cr_canseeotheruids(struct ucred *u1, struct ucred *u2)
{
if (!see_other_uids && u1->cr_ruid != u2->cr_ruid) {
if (priv_check_cred(u1, PRIV_SEEOTHERUIDS) != 0)
return (ESRCH);
}
return (0);
}
/*
* 'see_other_gids' determines whether or not visibility of processes
* and sockets with credentials holding different real gids is possible
* using a variety of system MIBs.
* XXX: data declarations should be together near the beginning of the file.
*/
static int see_other_gids = 1;
SYSCTL_INT(_security_bsd, OID_AUTO, see_other_gids, CTLFLAG_RW,
&see_other_gids, 0,
"Unprivileged processes may see subjects/objects with different real gid");
/*
* Determine if u1 can "see" the subject specified by u2, according to the
* 'see_other_gids' policy.
* Returns: 0 for permitted, ESRCH otherwise
* Locks: none
* References: *u1 and *u2 must not change during the call
* u1 may equal u2, in which case only one reference is required
*/
int
cr_canseeothergids(struct ucred *u1, struct ucred *u2)
{
int i, match;
if (!see_other_gids) {
match = 0;
for (i = 0; i < u1->cr_ngroups; i++) {
if (groupmember(u1->cr_groups[i], u2))
match = 1;
if (match)
break;
}
if (!match) {
if (priv_check_cred(u1, PRIV_SEEOTHERGIDS) != 0)
return (ESRCH);
}
}
return (0);
}
/*
* 'see_jail_proc' determines whether or not visibility of processes and
* sockets with credentials holding different jail ids is possible using a
* variety of system MIBs.
*
* XXX: data declarations should be together near the beginning of the file.
*/
static int see_jail_proc = 1;
SYSCTL_INT(_security_bsd, OID_AUTO, see_jail_proc, CTLFLAG_RW,
&see_jail_proc, 0,
"Unprivileged processes may see subjects/objects with different jail ids");
/*-
* Determine if u1 "can see" the subject specified by u2, according to the
* 'see_jail_proc' policy.
* Returns: 0 for permitted, ESRCH otherwise
* Locks: none
* References: *u1 and *u2 must not change during the call
* u1 may equal u2, in which case only one reference is required
*/
int
cr_canseejailproc(struct ucred *u1, struct ucred *u2)
{
if (u1->cr_uid == 0)
return (0);
return (!see_jail_proc && u1->cr_prison != u2->cr_prison ? ESRCH : 0);
}
/*-
* Determine if u1 "can see" the subject specified by u2.
* Returns: 0 for permitted, an errno value otherwise
* Locks: none
* References: *u1 and *u2 must not change during the call
* u1 may equal u2, in which case only one reference is required
*/
int
cr_cansee(struct ucred *u1, struct ucred *u2)
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
{
int error;
if ((error = prison_check(u1, u2)))
return (error);
#ifdef MAC
if ((error = mac_cred_check_visible(u1, u2)))
return (error);
#endif
if ((error = cr_canseeotheruids(u1, u2)))
return (error);
if ((error = cr_canseeothergids(u1, u2)))
return (error);
if ((error = cr_canseejailproc(u1, u2)))
return (error);
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
return (0);
}
/*-
* Determine if td "can see" the subject specified by p.
* Returns: 0 for permitted, an errno value otherwise
* Locks: Sufficient locks to protect p->p_ucred must be held. td really
* should be curthread.
* References: td and p must be valid for the lifetime of the call
*/
int
p_cansee(struct thread *td, struct proc *p)
{
/* Wrap cr_cansee() for all functionality. */
KASSERT(td == curthread, ("%s: td not curthread", __func__));
PROC_LOCK_ASSERT(p, MA_OWNED);
if (td->td_proc == p)
return (0);
return (cr_cansee(td->td_ucred, p->p_ucred));
}
/*
* 'conservative_signals' prevents the delivery of a broad class of
* signals by unprivileged processes to processes that have changed their
* credentials since the last invocation of execve(). This can prevent
* the leakage of cached information or retained privileges as a result
* of a common class of signal-related vulnerabilities. However, this
* may interfere with some applications that expect to be able to
* deliver these signals to peer processes after having given up
* privilege.
*/
static int conservative_signals = 1;
SYSCTL_INT(_security_bsd, OID_AUTO, conservative_signals, CTLFLAG_RW,
&conservative_signals, 0, "Unprivileged processes prevented from "
"sending certain signals to processes whose credentials have changed");
/*-
* Determine whether cred may deliver the specified signal to proc.
* Returns: 0 for permitted, an errno value otherwise.
* Locks: A lock must be held for proc.
* References: cred and proc must be valid for the lifetime of the call.
*/
int
cr_cansignal(struct ucred *cred, struct proc *proc, int signum)
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
{
int error;
2001-10-11 22:49:27 +00:00
PROC_LOCK_ASSERT(proc, MA_OWNED);
/*
* Jail semantics limit the scope of signalling to proc in the
* same jail as cred, if cred is in jail.
*/
error = prison_check(cred, proc->p_ucred);
if (error)
return (error);
#ifdef MAC
if ((error = mac_proc_check_signal(cred, proc, signum)))
return (error);
#endif
if ((error = cr_canseeotheruids(cred, proc->p_ucred)))
return (error);
if ((error = cr_canseeothergids(cred, proc->p_ucred)))
return (error);
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
/*
* UNIX signal semantics depend on the status of the P_SUGID
* bit on the target process. If the bit is set, then additional
* restrictions are placed on the set of available signals.
*/
if (conservative_signals && (proc->p_flag & P_SUGID)) {
switch (signum) {
case 0:
case SIGKILL:
case SIGINT:
case SIGTERM:
case SIGALRM:
case SIGSTOP:
case SIGTTIN:
case SIGTTOU:
case SIGTSTP:
case SIGHUP:
case SIGUSR1:
case SIGUSR2:
/*
* Generally, permit job and terminal control
* signals.
*/
break;
default:
/* Not permitted without privilege. */
error = priv_check_cred(cred, PRIV_SIGNAL_SUGID);
if (error)
return (error);
}
}
/*
* Generally, the target credential's ruid or svuid must match the
* subject credential's ruid or euid.
*/
if (cred->cr_ruid != proc->p_ucred->cr_ruid &&
cred->cr_ruid != proc->p_ucred->cr_svuid &&
cred->cr_uid != proc->p_ucred->cr_ruid &&
cred->cr_uid != proc->p_ucred->cr_svuid) {
error = priv_check_cred(cred, PRIV_SIGNAL_DIFFCRED);
if (error)
return (error);
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
}
return (0);
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
}
/*-
* Determine whether td may deliver the specified signal to p.
* Returns: 0 for permitted, an errno value otherwise
* Locks: Sufficient locks to protect various components of td and p
* must be held. td must be curthread, and a lock must be
* held for p.
* References: td and p must be valid for the lifetime of the call
*/
int
p_cansignal(struct thread *td, struct proc *p, int signum)
{
KASSERT(td == curthread, ("%s: td not curthread", __func__));
PROC_LOCK_ASSERT(p, MA_OWNED);
if (td->td_proc == p)
return (0);
/*
* UNIX signalling semantics require that processes in the same
* session always be able to deliver SIGCONT to one another,
* overriding the remaining protections.
*/
/* XXX: This will require an additional lock of some sort. */
if (signum == SIGCONT && td->td_proc->p_session == p->p_session)
return (0);
/*
* Some compat layers use SIGTHR and higher signals for
* communication between different kernel threads of the same
* process, so that they expect that it's always possible to
* deliver them, even for suid applications where cr_cansignal() can
* deny such ability for security consideration. It should be
* pretty safe to do since the only way to create two processes
* with the same p_leader is via rfork(2).
*/
if (td->td_proc->p_leader != NULL && signum >= SIGTHR &&
signum < SIGTHR + 4 && td->td_proc->p_leader == p->p_leader)
return (0);
return (cr_cansignal(td->td_ucred, p, signum));
}
/*-
* Determine whether td may reschedule p.
* Returns: 0 for permitted, an errno value otherwise
* Locks: Sufficient locks to protect various components of td and p
* must be held. td must be curthread, and a lock must
* be held for p.
* References: td and p must be valid for the lifetime of the call
*/
int
p_cansched(struct thread *td, struct proc *p)
{
int error;
KASSERT(td == curthread, ("%s: td not curthread", __func__));
PROC_LOCK_ASSERT(p, MA_OWNED);
if (td->td_proc == p)
return (0);
if ((error = prison_check(td->td_ucred, p->p_ucred)))
return (error);
#ifdef MAC
if ((error = mac_proc_check_sched(td->td_ucred, p)))
return (error);
#endif
if ((error = cr_canseeotheruids(td->td_ucred, p->p_ucred)))
return (error);
if ((error = cr_canseeothergids(td->td_ucred, p->p_ucred)))
return (error);
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
if (td->td_ucred->cr_ruid != p->p_ucred->cr_ruid &&
td->td_ucred->cr_uid != p->p_ucred->cr_ruid) {
error = priv_check(td, PRIV_SCHED_DIFFCRED);
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
if (error)
return (error);
}
return (0);
}
2018-11-27 17:51:50 +00:00
/*
* Handle getting or setting the prison's unprivileged_proc_debug
* value.
*/
static int
sysctl_unprivileged_proc_debug(SYSCTL_HANDLER_ARGS)
{
int error, val;
val = prison_allow(req->td->td_ucred, PR_ALLOW_UNPRIV_DEBUG);
2018-11-27 17:51:50 +00:00
error = sysctl_handle_int(oidp, &val, 0, req);
if (error != 0 || req->newptr == NULL)
return (error);
if (val != 0 && val != 1)
return (EINVAL);
prison_set_allow(req->td->td_ucred, PR_ALLOW_UNPRIV_DEBUG, val);
return (0);
2018-11-27 17:51:50 +00:00
}
/*
* The 'unprivileged_proc_debug' flag may be used to disable a variety of
* unprivileged inter-process debugging services, including some procfs
* functionality, ptrace(), and ktrace(). In the past, inter-process
* debugging has been involved in a variety of security problems, and sites
* not requiring the service might choose to disable it when hardening
* systems.
*/
2018-11-27 17:51:50 +00:00
SYSCTL_PROC(_security_bsd, OID_AUTO, unprivileged_proc_debug,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_PRISON | CTLFLAG_SECURE |
CTLFLAG_MPSAFE, 0, 0, sysctl_unprivileged_proc_debug, "I",
"Unprivileged processes may use process debugging facilities");
/*-
* Determine whether td may debug p.
* Returns: 0 for permitted, an errno value otherwise
* Locks: Sufficient locks to protect various components of td and p
* must be held. td must be curthread, and a lock must
* be held for p.
* References: td and p must be valid for the lifetime of the call
*/
int
p_candebug(struct thread *td, struct proc *p)
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
{
int error, grpsubset, i, uidsubset;
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
KASSERT(td == curthread, ("%s: td not curthread", __func__));
PROC_LOCK_ASSERT(p, MA_OWNED);
if (td->td_proc == p)
return (0);
if ((error = priv_check(td, PRIV_DEBUG_UNPRIV)))
return (error);
if ((error = prison_check(td->td_ucred, p->p_ucred)))
return (error);
#ifdef MAC
if ((error = mac_proc_check_debug(td->td_ucred, p)))
return (error);
#endif
if ((error = cr_canseeotheruids(td->td_ucred, p->p_ucred)))
return (error);
if ((error = cr_canseeothergids(td->td_ucred, p->p_ucred)))
return (error);
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
/*
* Is p's group set a subset of td's effective group set? This
* includes p's egid, group access list, rgid, and svgid.
*/
grpsubset = 1;
for (i = 0; i < p->p_ucred->cr_ngroups; i++) {
if (!groupmember(p->p_ucred->cr_groups[i], td->td_ucred)) {
grpsubset = 0;
break;
}
}
grpsubset = grpsubset &&
groupmember(p->p_ucred->cr_rgid, td->td_ucred) &&
groupmember(p->p_ucred->cr_svgid, td->td_ucred);
/*
* Are the uids present in p's credential equal to td's
* effective uid? This includes p's euid, svuid, and ruid.
*/
uidsubset = (td->td_ucred->cr_uid == p->p_ucred->cr_uid &&
td->td_ucred->cr_uid == p->p_ucred->cr_svuid &&
td->td_ucred->cr_uid == p->p_ucred->cr_ruid);
/*
* If p's gids aren't a subset, or the uids aren't a subset,
* or the credential has changed, require appropriate privilege
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
* for td to debug p.
*/
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
if (!grpsubset || !uidsubset) {
error = priv_check(td, PRIV_DEBUG_DIFFCRED);
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
if (error)
return (error);
}
/*
* Has the credential of the process changed since the last exec()?
*/
if ((p->p_flag & P_SUGID) != 0) {
error = priv_check(td, PRIV_DEBUG_SUGID);
if (error)
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
return (error);
}
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
/* Can't trace init when securelevel > 0. */
if (p == initproc) {
error = securelevel_gt(td->td_ucred, 0);
if (error)
return (error);
}
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
/*
* Can't trace a process that's currently exec'ing.
Add a new priv(9) kernel interface for checking the availability of privilege for threads and credentials. Unlike the existing suser(9) interface, priv(9) exposes a named privilege identifier to the privilege checking code, allowing more complex policies regarding the granting of privilege to be expressed. Two interfaces are provided, replacing the existing suser(9) interface: suser(td) -> priv_check(td, priv) suser_cred(cred, flags) -> priv_check_cred(cred, priv, flags) A comprehensive list of currently available kernel privileges may be found in priv.h. New privileges are easily added as required, but the comments on adding privileges found in priv.h and priv(9) should be read before doing so. The new privilege interface exposed sufficient information to the privilege checking routine that it will now be possible for jail to determine whether a particular privilege is granted in the check routine, rather than relying on hints from the calling context via the SUSER_ALLOWJAIL flag. For now, the flag is maintained, but a new jail check function, prison_priv_check(), is exposed from kern_jail.c and used by the privilege check routine to determine if the privilege is permitted in jail. As a result, a centralized list of privileges permitted in jail is now present in kern_jail.c. The MAC Framework is now also able to instrument privilege checks, both to deny privileges otherwise granted (mac_priv_check()), and to grant privileges otherwise denied (mac_priv_grant()), permitting MAC Policy modules to implement privilege models, as well as control a much broader range of system behavior in order to constrain processes running with root privilege. The suser() and suser_cred() functions remain implemented, now in terms of priv_check() and the PRIV_ROOT privilege, for use during the transition and possibly continuing use by third party kernel modules that have not been updated. The PRIV_DRIVER privilege exists to allow device drivers to check privilege without adopting a more specific privilege identifier. This change does not modify the actual security policy, rather, it modifies the interface for privilege checks so changes to the security policy become more feasible. Sponsored by: nCircle Network Security, Inc. Obtained from: TrustedBSD Project Discussed on: arch@ Reviewed (at least in part) by: mlaier, jmg, pjd, bde, ceri, Alex Lyashkov <umka at sevcity dot net>, Skip Ford <skip dot ford at verizon dot net>, Antoine Brodin <antoine dot brodin at laposte dot net>
2006-11-06 13:37:19 +00:00
*
* XXX: Note, this is not a security policy decision, it's a
* basic correctness/functionality decision. Therefore, this check
* should be moved to the caller's of p_candebug().
*/
if ((p->p_flag & P_INEXEC) != 0)
return (EBUSY);
/* Denied explicitly */
if ((p->p_flag2 & P2_NOTRACE) != 0) {
error = priv_check(td, PRIV_DEBUG_DENIED);
if (error != 0)
return (error);
}
o Centralize inter-process access control, introducing: int p_can(p1, p2, operation, privused) which allows specification of subject process, object process, inter-process operation, and an optional call-by-reference privused flag, allowing the caller to determine if privilege was required for the call to succeed. This allows jail, kern.ps_showallprocs and regular credential-based interaction checks to occur in one block of code. Possible operations are P_CAN_SEE, P_CAN_SCHED, P_CAN_KILL, and P_CAN_DEBUG. p_can currently breaks out as a wrapper to a series of static function checks in kern_prot, which should not be invoked directly. o Commented out capabilities entries are included for some checks. o Update most inter-process authorization to make use of p_can() instead of manual checks, PRISON_CHECK(), P_TRESPASS(), and kern.ps_showallprocs. o Modify suser{,_xxx} to use const arguments, as it no longer modifies process flags due to the disabling of ASU. o Modify some checks/errors in procfs so that ENOENT is returned instead of ESRCH, further improving concealment of processes that should not be visible to other processes. Also introduce new access checks to improve hiding of processes for procfs_lookup(), procfs_getattr(), procfs_readdir(). Correct a bug reported by bp concerning not handling the CREATE case in procfs_lookup(). Remove volatile flag in procfs that caused apparently spurious qualifier warnigns (approved by bde). o Add comment noting that ktrace() has not been updated, as its access control checks are different from ptrace(), whereas they should probably be the same. Further discussion should happen on this topic. Reviewed by: bde, green, phk, freebsd-security, others Approved by: bde Obtained from: TrustedBSD Project
2000-08-30 04:49:09 +00:00
return (0);
}
/*-
* Determine whether the subject represented by cred can "see" a socket.
* Returns: 0 for permitted, ENOENT otherwise.
*/
int
cr_canseesocket(struct ucred *cred, struct socket *so)
{
int error;
error = prison_check(cred, so->so_cred);
if (error)
return (ENOENT);
#ifdef MAC
error = mac_socket_check_visible(cred, so);
if (error)
return (error);
#endif
if (cr_canseeotheruids(cred, so->so_cred))
return (ENOENT);
if (cr_canseeothergids(cred, so->so_cred))
return (ENOENT);
return (0);
}
/*-
* Determine whether td can wait for the exit of p.
* Returns: 0 for permitted, an errno value otherwise
* Locks: Sufficient locks to protect various components of td and p
* must be held. td must be curthread, and a lock must
* be held for p.
* References: td and p must be valid for the lifetime of the call
*/
int
p_canwait(struct thread *td, struct proc *p)
{
int error;
KASSERT(td == curthread, ("%s: td not curthread", __func__));
PROC_LOCK_ASSERT(p, MA_OWNED);
if ((error = prison_check(td->td_ucred, p->p_ucred)))
return (error);
#ifdef MAC
if ((error = mac_proc_check_wait(td->td_ucred, p)))
return (error);
#endif
#if 0
/* XXXMAC: This could have odd effects on some shells. */
if ((error = cr_canseeotheruids(td->td_ucred, p->p_ucred)))
return (error);
#endif
return (0);
}
/*
* Credential management.
*
* struct ucred objects are rarely allocated but gain and lose references all
* the time (e.g., on struct file alloc/dealloc) turning refcount updates into
* a significant source of cache-line ping ponging. Common cases are worked
* around by modifying thread-local counter instead if the cred to operate on
* matches td_realucred.
*
* The counter is split into 2 parts:
* - cr_users -- total count of all struct proc and struct thread objects
* which have given cred in p_ucred and td_ucred respectively
* - cr_ref -- the actual ref count, only valid if cr_users == 0
*
* If users == 0 then cr_ref behaves similarly to refcount(9), in particular if
* the count reaches 0 the object is freeable.
* If users > 0 and curthread->td_realucred == cred, then updates are performed
* against td_ucredref.
* In other cases updates are performed against cr_ref.
*
* Changing td_realucred into something else decrements cr_users and transfers
* accumulated updates.
*/
struct ucred *
crcowget(struct ucred *cr)
{
mtx_lock(&cr->cr_mtx);
KASSERT(cr->cr_users > 0, ("%s: users %d not > 0 on cred %p",
__func__, cr->cr_users, cr));
cr->cr_users++;
cr->cr_ref++;
mtx_unlock(&cr->cr_mtx);
return (cr);
}
static struct ucred *
crunuse(struct thread *td)
{
struct ucred *cr, *crold;
MPASS(td->td_realucred == td->td_ucred);
cr = td->td_realucred;
mtx_lock(&cr->cr_mtx);
cr->cr_ref += td->td_ucredref;
td->td_ucredref = 0;
KASSERT(cr->cr_users > 0, ("%s: users %d not > 0 on cred %p",
__func__, cr->cr_users, cr));
cr->cr_users--;
if (cr->cr_users == 0) {
KASSERT(cr->cr_ref > 0, ("%s: ref %ld not > 0 on cred %p",
__func__, cr->cr_ref, cr));
crold = cr;
} else {
cr->cr_ref--;
crold = NULL;
}
mtx_unlock(&cr->cr_mtx);
td->td_realucred = NULL;
return (crold);
}
2020-11-14 19:22:02 +00:00
static void
crunusebatch(struct ucred *cr, int users, int ref)
{
KASSERT(users > 0, ("%s: passed users %d not > 0 ; cred %p",
__func__, users, cr));
mtx_lock(&cr->cr_mtx);
KASSERT(cr->cr_users >= users, ("%s: users %d not > %d on cred %p",
__func__, cr->cr_users, users, cr));
cr->cr_users -= users;
cr->cr_ref += ref;
cr->cr_ref -= users;
if (cr->cr_users > 0) {
mtx_unlock(&cr->cr_mtx);
return;
}
KASSERT(cr->cr_ref >= 0, ("%s: ref %ld not >= 0 on cred %p",
2020-11-14 19:22:02 +00:00
__func__, cr->cr_ref, cr));
if (cr->cr_ref > 0) {
mtx_unlock(&cr->cr_mtx);
return;
}
crfree_final(cr);
}
void
crcowfree(struct thread *td)
{
struct ucred *cr;
cr = crunuse(td);
if (cr != NULL)
crfree(cr);
}
struct ucred *
crcowsync(void)
{
struct thread *td;
struct proc *p;
struct ucred *crnew, *crold;
td = curthread;
p = td->td_proc;
PROC_LOCK_ASSERT(p, MA_OWNED);
MPASS(td->td_realucred == td->td_ucred);
if (td->td_realucred == p->p_ucred)
return (NULL);
crnew = crcowget(p->p_ucred);
crold = crunuse(td);
td->td_realucred = crnew;
td->td_ucred = td->td_realucred;
return (crold);
}
2020-11-14 19:22:02 +00:00
/*
* Batching.
*/
void
credbatch_add(struct credbatch *crb, struct thread *td)
{
struct ucred *cr;
MPASS(td->td_realucred != NULL);
MPASS(td->td_realucred == td->td_ucred);
MPASS(TD_GET_STATE(td) == TDS_INACTIVE);
2020-11-14 19:22:02 +00:00
cr = td->td_realucred;
KASSERT(cr->cr_users > 0, ("%s: users %d not > 0 on cred %p",
__func__, cr->cr_users, cr));
if (crb->cred != cr) {
if (crb->users > 0) {
MPASS(crb->cred != NULL);
crunusebatch(crb->cred, crb->users, crb->ref);
crb->users = 0;
crb->ref = 0;
}
}
crb->cred = cr;
crb->users++;
crb->ref += td->td_ucredref;
td->td_ucredref = 0;
td->td_realucred = NULL;
}
void
credbatch_final(struct credbatch *crb)
{
MPASS(crb->cred != NULL);
MPASS(crb->users > 0);
crunusebatch(crb->cred, crb->users, crb->ref);
}
1994-05-24 10:09:53 +00:00
/*
* Allocate a zeroed cred structure.
*/
struct ucred *
crget(void)
1994-05-24 10:09:53 +00:00
{
struct ucred *cr;
1994-05-24 10:09:53 +00:00
cr = malloc(sizeof(*cr), M_CRED, M_WAITOK | M_ZERO);
mtx_init(&cr->cr_mtx, "cred", NULL, MTX_DEF);
cr->cr_ref = 1;
#ifdef AUDIT
audit_cred_init(cr);
#endif
#ifdef MAC
mac_cred_init(cr);
#endif
cr->cr_groups = cr->cr_smallgroups;
cr->cr_agroups =
sizeof(cr->cr_smallgroups) / sizeof(cr->cr_smallgroups[0]);
1994-05-24 10:09:53 +00:00
return (cr);
}
/*
* Claim another reference to a ucred structure.
*/
struct ucred *
crhold(struct ucred *cr)
{
struct thread *td;
td = curthread;
if (__predict_true(td->td_realucred == cr)) {
KASSERT(cr->cr_users > 0, ("%s: users %d not > 0 on cred %p",
__func__, cr->cr_users, cr));
td->td_ucredref++;
return (cr);
}
mtx_lock(&cr->cr_mtx);
cr->cr_ref++;
mtx_unlock(&cr->cr_mtx);
return (cr);
}
1994-05-24 10:09:53 +00:00
/*
* Free a cred structure. Throws away space when ref count gets to 0.
1994-05-24 10:09:53 +00:00
*/
void
crfree(struct ucred *cr)
1994-05-24 10:09:53 +00:00
{
struct thread *td;
td = curthread;
if (__predict_true(td->td_realucred == cr)) {
KASSERT(cr->cr_users > 0, ("%s: users %d not > 0 on cred %p",
__func__, cr->cr_users, cr));
td->td_ucredref--;
return;
}
mtx_lock(&cr->cr_mtx);
KASSERT(cr->cr_users >= 0, ("%s: users %d not >= 0 on cred %p",
__func__, cr->cr_users, cr));
cr->cr_ref--;
if (cr->cr_users > 0) {
mtx_unlock(&cr->cr_mtx);
return;
}
KASSERT(cr->cr_ref >= 0, ("%s: ref %ld not >= 0 on cred %p",
__func__, cr->cr_ref, cr));
if (cr->cr_ref > 0) {
mtx_unlock(&cr->cr_mtx);
return;
}
2020-11-14 19:22:02 +00:00
crfree_final(cr);
}
static void
crfree_final(struct ucred *cr)
{
KASSERT(cr->cr_users == 0, ("%s: users %d not == 0 on cred %p",
__func__, cr->cr_users, cr));
KASSERT(cr->cr_ref == 0, ("%s: ref %ld not == 0 on cred %p",
2020-11-14 19:22:02 +00:00
__func__, cr->cr_ref, cr));
/*
* Some callers of crget(), such as nfs_statfs(), allocate a temporary
* credential, but don't allocate a uidinfo structure.
*/
if (cr->cr_uidinfo != NULL)
uifree(cr->cr_uidinfo);
if (cr->cr_ruidinfo != NULL)
uifree(cr->cr_ruidinfo);
if (cr->cr_prison != NULL)
prison_free(cr->cr_prison);
if (cr->cr_loginclass != NULL)
loginclass_free(cr->cr_loginclass);
#ifdef AUDIT
audit_cred_destroy(cr);
#endif
#ifdef MAC
mac_cred_destroy(cr);
#endif
mtx_destroy(&cr->cr_mtx);
if (cr->cr_groups != cr->cr_smallgroups)
free(cr->cr_groups, M_CRED);
free(cr, M_CRED);
1994-05-24 10:09:53 +00:00
}
/*
* Copy a ucred's contents from a template. Does not block.
*/
void
crcopy(struct ucred *dest, struct ucred *src)
{
2014-10-05 02:16:53 +00:00
KASSERT(dest->cr_ref == 1, ("crcopy of shared ucred"));
bcopy(&src->cr_startcopy, &dest->cr_startcopy,
(unsigned)((caddr_t)&src->cr_endcopy -
(caddr_t)&src->cr_startcopy));
dest->cr_flags = src->cr_flags;
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
crsetgroups(dest, src->cr_ngroups, src->cr_groups);
uihold(dest->cr_uidinfo);
uihold(dest->cr_ruidinfo);
prison_hold(dest->cr_prison);
loginclass_hold(dest->cr_loginclass);
#ifdef AUDIT
audit_cred_copy(src, dest);
#endif
#ifdef MAC
mac_cred_copy(src, dest);
#endif
1994-05-24 10:09:53 +00:00
}
/*
* Dup cred struct to a new held one.
*/
struct ucred *
crdup(struct ucred *cr)
1994-05-24 10:09:53 +00:00
{
struct ucred *newcr;
newcr = crget();
crcopy(newcr, cr);
1994-05-24 10:09:53 +00:00
return (newcr);
}
/*
* Fill in a struct xucred based on a struct ucred.
*/
void
cru2x(struct ucred *cr, struct xucred *xcr)
{
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
int ngroups;
bzero(xcr, sizeof(*xcr));
xcr->cr_version = XUCRED_VERSION;
xcr->cr_uid = cr->cr_uid;
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
ngroups = MIN(cr->cr_ngroups, XU_NGROUPS);
xcr->cr_ngroups = ngroups;
bcopy(cr->cr_groups, xcr->cr_groups,
ngroups * sizeof(*cr->cr_groups));
}
void
cru2xt(struct thread *td, struct xucred *xcr)
{
cru2x(td->td_ucred, xcr);
xcr->cr_pid = td->td_proc->p_pid;
}
/*
* Set initial process credentials.
* Callers are responsible for providing the reference for provided credentials.
*/
void
proc_set_cred_init(struct proc *p, struct ucred *newcred)
{
p->p_ucred = crcowget(newcred);
}
/*
* Change process credentials.
* Callers are responsible for providing the reference for passed credentials
* and for freeing old ones.
*
* Process has to be locked except when it does not have credentials (as it
* should not be visible just yet) or when newcred is NULL (as this can be
* only used when the process is about to be freed, at which point it should
* not be visible anymore).
*/
void
proc_set_cred(struct proc *p, struct ucred *newcred)
{
struct ucred *cr;
cr = p->p_ucred;
MPASS(cr != NULL);
PROC_LOCK_ASSERT(p, MA_OWNED);
KASSERT(newcred->cr_users == 0, ("%s: users %d not 0 on cred %p",
__func__, newcred->cr_users, newcred));
mtx_lock(&cr->cr_mtx);
KASSERT(cr->cr_users > 0, ("%s: users %d not > 0 on cred %p",
__func__, cr->cr_users, cr));
cr->cr_users--;
mtx_unlock(&cr->cr_mtx);
p->p_ucred = newcred;
newcred->cr_users = 1;
PROC_UPDATE_COW(p);
}
void
proc_unset_cred(struct proc *p)
{
struct ucred *cr;
MPASS(p->p_state == PRS_ZOMBIE || p->p_state == PRS_NEW);
cr = p->p_ucred;
p->p_ucred = NULL;
KASSERT(cr->cr_users > 0, ("%s: users %d not > 0 on cred %p",
__func__, cr->cr_users, cr));
mtx_lock(&cr->cr_mtx);
cr->cr_users--;
if (cr->cr_users == 0)
KASSERT(cr->cr_ref > 0, ("%s: ref %ld not > 0 on cred %p",
__func__, cr->cr_ref, cr));
mtx_unlock(&cr->cr_mtx);
crfree(cr);
}
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
struct ucred *
crcopysafe(struct proc *p, struct ucred *cr)
{
struct ucred *oldcred;
int groups;
PROC_LOCK_ASSERT(p, MA_OWNED);
oldcred = p->p_ucred;
while (cr->cr_agroups < oldcred->cr_agroups) {
groups = oldcred->cr_agroups;
PROC_UNLOCK(p);
crextend(cr, groups);
PROC_LOCK(p);
oldcred = p->p_ucred;
}
crcopy(cr, oldcred);
return (oldcred);
}
/*
* Extend the passed in credential to hold n items.
*/
void
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
crextend(struct ucred *cr, int n)
{
int cnt;
/* Truncate? */
if (n <= cr->cr_agroups)
return;
/*
* We extend by 2 each time since we're using a power of two
* allocator until we need enough groups to fill a page.
* Once we're allocating multiple pages, only allocate as many
* as we actually need. The case of processes needing a
* non-power of two number of pages seems more likely than
* a real world process that adds thousands of groups one at a
* time.
*/
if ( n < PAGE_SIZE / sizeof(gid_t) ) {
if (cr->cr_agroups == 0)
cnt = MAX(1, MINALLOCSIZE / sizeof(gid_t));
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
else
cnt = cr->cr_agroups * 2;
while (cnt < n)
cnt *= 2;
} else
cnt = roundup2(n, PAGE_SIZE / sizeof(gid_t));
/* Free the old array. */
if (cr->cr_groups != cr->cr_smallgroups)
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
free(cr->cr_groups, M_CRED);
cr->cr_groups = malloc(cnt * sizeof(gid_t), M_CRED, M_WAITOK | M_ZERO);
cr->cr_agroups = cnt;
}
/*
* Copy groups in to a credential, preserving any necessary invariants.
* Currently this includes the sorting of all supplemental gids.
* crextend() must have been called before hand to ensure sufficient
* space is available.
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
*/
static void
crsetgroups_locked(struct ucred *cr, int ngrp, gid_t *groups)
{
int i;
int j;
gid_t g;
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
KASSERT(cr->cr_agroups >= ngrp, ("cr_ngroups is too small"));
bcopy(groups, cr->cr_groups, ngrp * sizeof(gid_t));
cr->cr_ngroups = ngrp;
/*
* Sort all groups except cr_groups[0] to allow groupmember to
* perform a binary search.
*
* XXX: If large numbers of groups become common this should
* be replaced with shell sort like linux uses or possibly
* heap sort.
*/
for (i = 2; i < ngrp; i++) {
g = cr->cr_groups[i];
for (j = i-1; j >= 1 && g < cr->cr_groups[j]; j--)
cr->cr_groups[j + 1] = cr->cr_groups[j];
cr->cr_groups[j + 1] = g;
}
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
}
/*
* Copy groups in to a credential after expanding it if required.
* Truncate the list to (ngroups_max + 1) if it is too large.
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
*/
void
crsetgroups(struct ucred *cr, int ngrp, gid_t *groups)
{
if (ngrp > ngroups_max + 1)
ngrp = ngroups_max + 1;
Rework the credential code to support larger values of NGROUPS and NGROUPS_MAX, eliminate ABI dependencies on them, and raise the to 1024 and 1023 respectively. (Previously they were equal, but under a close reading of POSIX, NGROUPS_MAX was defined to be too large by 1 since it is the number of supplemental groups, not total number of groups.) The bulk of the change consists of converting the struct ucred member cr_groups from a static array to a pointer. Do the equivalent in kinfo_proc. Introduce new interfaces crcopysafe() and crsetgroups() for duplicating a process credential before modifying it and for setting group lists respectively. Both interfaces take care for the details of allocating groups array. crsetgroups() takes care of truncating the group list to the current maximum (NGROUPS) if necessary. In the future, crsetgroups() may be responsible for insuring invariants such as sorting the supplemental groups to allow groupmember() to be implemented as a binary search. Because we can not change struct xucred without breaking application ABIs, we leave it alone and introduce a new XU_NGROUPS value which is always 16 and is to be used or NGRPS as appropriate for things such as NFS which need to use no more than 16 groups. When feasible, truncate the group list rather than generating an error. Minor changes: - Reduce the number of hand rolled versions of groupmember(). - Do not assign to both cr_gid and cr_groups[0]. - Modify ipfw to cache ucreds instead of part of their contents since they are immutable once referenced by more than one entity. Submitted by: Isilon Systems (initial implementation) X-MFC after: never PR: bin/113398 kern/133867
2009-06-19 17:10:35 +00:00
crextend(cr, ngrp);
crsetgroups_locked(cr, ngrp, groups);
}
1994-05-24 10:09:53 +00:00
/*
* Get login name, if available.
*/
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct getlogin_args {
char *namebuf;
u_int namelen;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_getlogin(struct thread *td, struct getlogin_args *uap)
1994-05-24 10:09:53 +00:00
{
char login[MAXLOGNAME];
struct proc *p = td->td_proc;
size_t len;
1994-05-24 10:09:53 +00:00
if (uap->namelen > MAXLOGNAME)
1997-03-03 23:02:43 +00:00
uap->namelen = MAXLOGNAME;
PROC_LOCK(p);
SESS_LOCK(p->p_session);
len = strlcpy(login, p->p_session->s_login, uap->namelen) + 1;
SESS_UNLOCK(p->p_session);
PROC_UNLOCK(p);
if (len > uap->namelen)
return (ERANGE);
return (copyout(login, uap->namebuf, len));
1994-05-24 10:09:53 +00:00
}
/*
* Set login name.
*/
#ifndef _SYS_SYSPROTO_H_
1994-05-24 10:09:53 +00:00
struct setlogin_args {
char *namebuf;
};
#endif
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
int
sys_setlogin(struct thread *td, struct setlogin_args *uap)
1994-05-24 10:09:53 +00:00
{
struct proc *p = td->td_proc;
1994-05-24 10:09:53 +00:00
int error;
char logintmp[MAXLOGNAME];
1994-05-24 10:09:53 +00:00
CTASSERT(sizeof(p->p_session->s_login) >= sizeof(logintmp));
error = priv_check(td, PRIV_PROC_SETLOGIN);
if (error)
return (error);
error = copyinstr(uap->namebuf, logintmp, sizeof(logintmp), NULL);
if (error != 0) {
if (error == ENAMETOOLONG)
error = EINVAL;
return (error);
}
AUDIT_ARG_LOGIN(logintmp);
PROC_LOCK(p);
SESS_LOCK(p->p_session);
strcpy(p->p_session->s_login, logintmp);
SESS_UNLOCK(p->p_session);
PROC_UNLOCK(p);
return (0);
1994-05-24 10:09:53 +00:00
}
void
setsugid(struct proc *p)
{
PROC_LOCK_ASSERT(p, MA_OWNED);
p->p_flag |= P_SUGID;
}
/*-
* Change a process's effective uid.
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
* Side effects: newcred->cr_uid and newcred->cr_uidinfo will be modified.
* References: newcred must be an exclusive credential reference for the
* duration of the call.
*/
void
change_euid(struct ucred *newcred, struct uidinfo *euip)
{
newcred->cr_uid = euip->ui_uid;
uihold(euip);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
uifree(newcred->cr_uidinfo);
newcred->cr_uidinfo = euip;
}
/*-
* Change a process's effective gid.
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
* Side effects: newcred->cr_gid will be modified.
* References: newcred must be an exclusive credential reference for the
* duration of the call.
*/
void
change_egid(struct ucred *newcred, gid_t egid)
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
{
newcred->cr_groups[0] = egid;
}
/*-
* Change a process's real uid.
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
* Side effects: newcred->cr_ruid will be updated, newcred->cr_ruidinfo
* will be updated, and the old and new cr_ruidinfo proc
* counts will be updated.
* References: newcred must be an exclusive credential reference for the
* duration of the call.
*/
void
change_ruid(struct ucred *newcred, struct uidinfo *ruip)
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
{
(void)chgproccnt(newcred->cr_ruidinfo, -1, 0);
newcred->cr_ruid = ruip->ui_uid;
uihold(ruip);
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
uifree(newcred->cr_ruidinfo);
newcred->cr_ruidinfo = ruip;
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
(void)chgproccnt(newcred->cr_ruidinfo, 1, 0);
}
/*-
* Change a process's real gid.
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
* Side effects: newcred->cr_rgid will be updated.
* References: newcred must be an exclusive credential reference for the
* duration of the call.
*/
void
change_rgid(struct ucred *newcred, gid_t rgid)
{
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
newcred->cr_rgid = rgid;
}
/*-
* Change a process's saved uid.
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
* Side effects: newcred->cr_svuid will be updated.
* References: newcred must be an exclusive credential reference for the
* duration of the call.
*/
void
change_svuid(struct ucred *newcred, uid_t svuid)
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
{
newcred->cr_svuid = svuid;
}
/*-
* Change a process's saved gid.
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
* Side effects: newcred->cr_svgid will be updated.
* References: newcred must be an exclusive credential reference for the
* duration of the call.
*/
void
change_svgid(struct ucred *newcred, gid_t svgid)
o Merge contents of struct pcred into struct ucred. Specifically, add the real uid, saved uid, real gid, and saved gid to ucred, as well as the pcred->pc_uidinfo, which was associated with the real uid, only rename it to cr_ruidinfo so as not to conflict with cr_uidinfo, which corresponds to the effective uid. o Remove p_cred from struct proc; add p_ucred to struct proc, replacing original macro that pointed. p->p_ucred to p->p_cred->pc_ucred. o Universally update code so that it makes use of ucred instead of pcred, p->p_ucred instead of p->p_pcred, cr_ruidinfo instead of p_uidinfo, cr_{r,sv}{u,g}id instead of p_*, etc. o Remove pcred0 and its initialization from init_main.c; initialize cr_ruidinfo there. o Restruction many credential modification chunks to always crdup while we figure out locking and optimizations; generally speaking, this means moving to a structure like this: newcred = crdup(oldcred); ... p->p_ucred = newcred; crfree(oldcred); It's not race-free, but better than nothing. There are also races in sys_process.c, all inter-process authorization, fork, exec, and exit. o Remove sigio->sio_ruid since sigio->sio_ucred now contains the ruid; remove comments indicating that the old arrangement was a problem. o Restructure exec1() a little to use newcred/oldcred arrangement, and use improved uid management primitives. o Clean up exit1() so as to do less work in credential cleanup due to pcred removal. o Clean up fork1() so as to do less work in credential cleanup and allocation. o Clean up ktrcanset() to take into account changes, and move to using suser_xxx() instead of performing a direct uid==0 comparision. o Improve commenting in various kern_prot.c credential modification calls to better document current behavior. In a couple of places, current behavior is a little questionable and we need to check POSIX.1 to make sure it's "right". More commenting work still remains to be done. o Update credential management calls, such as crfree(), to take into account new ruidinfo reference. o Modify or add the following uid and gid helper routines: change_euid() change_egid() change_ruid() change_rgid() change_svuid() change_svgid() In each case, the call now acts on a credential not a process, and as such no longer requires more complicated process locking/etc. They now assume the caller will do any necessary allocation of an exclusive credential reference. Each is commented to document its reference requirements. o CANSIGIO() is simplified to require only credentials, not processes and pcreds. o Remove lots of (p_pcred==NULL) checks. o Add an XXX to authorization code in nfs_lock.c, since it's questionable, and needs to be considered carefully. o Simplify posix4 authorization code to require only credentials, not processes and pcreds. Note that this authorization, as well as CANSIGIO(), needs to be updated to use the p_cansignal() and p_cansched() centralized authorization routines, as they currently do not take into account some desirable restrictions that are handled by the centralized routines, as well as being inconsistent with other similar authorization instances. o Update libkvm to take these changes into account. Obtained from: TrustedBSD Project Reviewed by: green, bde, jhb, freebsd-arch, freebsd-audit
2001-05-25 16:59:11 +00:00
{
newcred->cr_svgid = svgid;
}
bool allow_ptrace = true;
SYSCTL_BOOL(_security_bsd, OID_AUTO, allow_ptrace, CTLFLAG_RWTUN,
&allow_ptrace, 0,
"Deny ptrace(2) use by returning ENOSYS");